通过NPU进行核函数的运行验证时,出现挂死现象;通过CPU进行核函数的运行验证时,出现AllocTensor/FreeTensor失败的报错,日志报错和调用栈打印如下:
1 2 3 4 5 6 7 8 9 10 | [ERROR][Core_0][/usr/local/Ascend/latest/x86_64-linux/tikcpp/tikcfw/interface/kernel_tpipe.h:730][AllocEventID][321678] current size is 4, max buffer number in same queue position is 4 [ERROR][CORE_0][pid 321674] error happened! ========= SIGABRT Signal (Abort Signal from abort) catched, backtrace info: [#0] 0x000000000001e7c0: handler(int) at /usr/local/Ascend/latest/tools/tikicpulib/lib/include/kern_fwk.h:105 [#1] 0x0000000000017c4f: signed char AscendC::TPipe::AllocEventID<(AscendC::HardEvent)5>() at /usr/local/Ascend/latest/x86_64-linux/tikcpp/tikcfw/interface/kernel_tpipe.h:733 [#2] 0x000000000001426d: AscendC::TQueBind<(AscendC::TPosition)0, (AscendC::TPosition)9, 4, 0>::FreeBuffer(unsigned char*) at /usr/local/Ascend/latest/x86_64-linux/tikcpp/tikcfw/interface/kernel_tpipe.h:1217 [#3] 0x0000000000011058: void AscendC::TQueBind<(AscendC::TPosition)0, (AscendC::TPosition)9, 4, 0>::FreeTensor<float16::Fp16T>(AscendC::LocalTensor<float16::Fp16T>&) at /usr/local/Ascend/latest/x86_64-linux/tikcpp/tikcfw/interface/kernel_tpipe.h:1237 [#4] 0x000000000000dfde: KernelAdd::Compute(int) at /home/xxxx/xxxx.cpp:59 [#5] 0x000000000000dd1c: KernelAdd::Process() at /home/xxxx/xxxx.cpp:37 (discriminator 2) ... |
根据日志信息“current size is 4, max buffer number in same queue position is 4”可以明确该问题是因为同一个TPosition上QUE Buffer的数量超出限制导致。
同一个TPosition上的所有Queue,连续调用AllocTensor接口申请的Tensor数量,根据AI处理器型号的不同,有数量约束。申请Buffer时,需要满足该约束。
不满足该约束,在后续使用AllocTensor/FreeTensor可能会出现分配资源失败。比如:
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 | AscendC::TQue<AscendC::TPosition::VECIN, 1> que0; AscendC::TQue<AscendC::TPosition::VECIN, 1> que1; AscendC::TQue<AscendC::TPosition::VECIN, 1> que2; AscendC::TQue<AscendC::TPosition::VECIN, 1> que3; AscendC::TQue<AscendC::TPosition::VECIN, 1> que4; AscendC::TQue<AscendC::TPosition::VECIN, 1> que5; // 比如,算子有6个输入,需要申请6块buffer // 通过6个队列为其申请内存,分别为que0~que5,每个que分配1块,申请VECIN TPosition上的buffer总数为6 // 假设,同一个Position上连续Alloc的Buffer数量限制为4,超出该限制后,使用AllocTensor/FreeTensor会出现分配资源失败 // 在NPU上可能体现为卡死等异常行为,在CPU Debug场景会出现报错提示 pipe.InitBuffer(que0, 1, len); pipe.InitBuffer(que1, 1, len); pipe.InitBuffer(que2, 1, len); pipe.InitBuffer(que3, 1, len); pipe.InitBuffer(que4, 1, len); pipe.InitBuffer(que5, 1, len); AscendC::LocalTensor<T> local1 = que0.AllocTensor<T>(); AscendC::LocalTensor<T> local2 = que1.AllocTensor<T>(); AscendC::LocalTensor<T> local3 = que2.AllocTensor<T>(); AscendC::LocalTensor<T> local4 = que3.AllocTensor<T>(); // 第5个AllocTensor会出现资源分配失败,同一个TPosition上同时Alloc出来的Tensor数量超出了4个的限制 AscendC::LocalTensor<T> local5 = que4.AllocTensor<T>(); |
如果确实有多块buffer使用,可以将多个buffer合并到一块buffer,通过偏移使用。样例如下:
1 2 3 4 5 6 7 8 9 10 11 12 13 14 | // 此时建议通过以下方法解决: // 如果确实有多块buffer使用, 可以将多个buffer合并到一块buffer, 通过偏移使用 pipe.InitBuffer(que0, 1, len * 3); pipe.InitBuffer(que1, 1, len * 3); /* * 分配出3块内存大小的LocalTensor, local1的地址为que0中buffer的起始地址, * local2的地址为local1的地址偏移len后的地址,local3的地址为local1的地址偏移 * len * 2的地址 */ int32_t offset1 = len; int32_t offset2 = len * 2; AscendC::LocalTensor<T> local1 = que0.AllocTensor<T>(); AscendC::LocalTensor<T> local2 = local1[offset1]; AscendC::LocalTensor<T> local3 = local1[offset2]; |